Hacking Diabetes: The DIY Bionic Pancreas Is Here to Stay
A do-it-yourself option to replace traditional diabetes management technology is growing in popularity.
Ask any patient with type 1 diabetes (T1D) about disease management, and you’ll likely hear stories of meticulous monitoring, hardware challenges, and more. So it’s no surprise that enterprising patients and caregivers have begun to hack their insulin pumps and continuous glucose monitors (CGMs) to create hybrid, closed-loop automated insulin delivery (AID) systems using off- the-shelf hardware and do-it-yourself (DIY), open-source smartphone apps to build their own bionic pancreas.
Aaron J. Kowalski, PhD, CEO of the Juvenile Diabetes Research Foundation (JDRF), received a diagnosis of T1D at age 13. When he learned about the DIY closed-loop AID system called Loop,1 he jumped right in. “I was a very early adopter because my first responsibility is to my health and my family. I had worked in the science for many years, and I felt that the risk was worth taking,” Kowalski said.
Kowalski isn’t alone. Many JDRF staffers have T1D, and they often use open-source DIY AID systems. The same is true at Integrated Diabetes Services. Owner and clinical director Gary Scheiner, MS, CDCES, and many of his staff clinicians use open- source AID.
“I wouldn’t consider myself tech savvy; I’m tech comfortable,” Scheiner said. “DIY is not much different conceptually from what’s available commercially today, but DIY systems like Loop are more flexible [and] more customizable. Commercially available systems have serious guardrails [and] very few variables that the user can adjust, which limits functionality. With something like Loop, you can achieve higher time in range, tighter averages, and lower [hemoglobin A levels] than with commercially available systems because you can adjust to your daily needs more easily.”
How It Started
Insulin pumps and CGMs have been a staple in T1D care since the 1980s and early 2000s, respectively. But for a time, the 2 devices could not communicate, as no feedback loop existed between CGMs and insulin pumps to automatically regulate insulin dosing and blood glucose levels.
In 2014, some insulin pump users realized that a security flaw in the now-discontinued Medtronic pump models let them hack their pumps. Anyone with the right tech savvy could create their own algorithm, load it on a smartphone, and use real-time CGM data to calculate insulin dosing and control their pump. And thus looping—closing the CGM-insulin pump loop using DIY, open-source soft- ware—was born.
Many early adopters were parents of children with T1D who wanted to be able to remotely monitor their child’s serum glucose and adjust insulin dosing as needed, Kowalski explained. Under the banner #WeAreNotWaiting, parents formed an online community, Nightscout,2 an “open-source DIY project that allows real time access to CGM data,” he said. In its first iteration, Nightscout was specifically made to remotely monitor Dexcom G4 CGM data. Eager users shared Open Artificial Pancreas System (OpenAPS)3 documentation, including open-source reference designs, safety information, and step-by-step implementation instructions. Everything was designed touse existing FDA-approved devices, commodity hardware, and open- source software. As the information spread, AndroidAPS (for Android devices) and Loop (for iPhones and later iOS devices) focused on users in device-specific communities.
By 2016, OpenAPS users were presenting data at scientific gatherings sponsored by the American Diabetes Association4 and other medical organizations focused on diabetes research and care. Tidepool added Loop data in 20185 and began developing a commercial algorithm with financial support from JDRF and other sources. In 2021, Tidepool completed the first open-source AID algorithm 510(k) application to the FDA.6,7
Tidepool forked the original DIY Loop to create a more efficient, easier to use, and better documented app for the FDA submission, said Christopher Snider, Tidepool’s community and clinic success manager. DIY loopers have incorporated some of the updates, but Tidepool remains a dis- tinct algorithm.
“We put a tight code base on the app and used human factors approaches to increase the safety and efficacy of the algorithm and improve the overall user experience,” Snider said. “When Tidepool becomes commercially available, we will have a support team in place to work with [individuals] directly, just as you would expect with any other insulin delivery system. Support is a little more nebulous in the DIY space where you go to Facebook or Twitter with questions.”
Keep Pharmacists in the Loop
In the medical device world, customer support is generally device specific; every pump, CGM, and hybrid closed- loop system has its own support team. Open source, though, lets users mix and match pumps and CGMs from different manufacturers, creating more flexibility and more choices in features.
“Insulin is prescribed as usual and the health care team still needs to figure out how much insulin the patient is using,” said Diana Isaacs, PharmD, BCPS, BCACP, BC-ADM, CDCES, endocrine clinical pharmacy specialist and remote monitoring program coordinator at the Cleveland Clinic Endocrinology and Metabolism Institute in Ohio. “With systems like Tidepool and Nightscout, we can view all the user data to figure insulin needs.”
She added, “I can’t configure a DIY system because it’s not FDA approved. But what I can do as a pharmacist is review patients’ data and advise them on ways to remain safe. We have a professional responsibility to support our patients and keep them safe.”
Although Isaacs sees open-source AID due to her location and her high visibility within the diabetes com- munity, data published in a recent international consensus statement in Lancet Diabetes and Endocrinology estimated that approximately 10,000 individuals with diabetes are using DIY AID—a miniscule figure considering the 9 million individuals world- wide who have T1D.8
John Beckner, RPh, senior director of strategic initiatives at the National Community Pharmacy Association, said that for now DIY is “just not that prevalent. And as technology continues to advance, there isn’t going to be much need to do it.”
The future of DIY will depend largely on industry. Dexcom, Medtronic, Tandem, Insulet, and other manufacturers regularly iterate existing products, bringing new systems online. And for now these products remain a closed ecosystem. “The tendency in industry is to want it all under your own shop,” Kowalski said. “If we fast forward 5 years, it would be shocking to me if we are not plugging and playing most of these com- ponents. We don’t have plug-and-play systems in the United States but we do in Europe.”9
European regulators approved CamAPS FX, an Android app created by CamDiab, a JDRF-funded project at Cambridge University, in 2020. The app, which links pumps and CGMs across manufacturers, is approved for systems used by patients 1 year and older.10
Kowalski predicted a future with fully interoperable closed-loop systems that combine pump X, sensor Y, and app Z. If industry doesn’t do it, DIYers will do it for them. “I was just talking with some of our team about when the Omnipod system is going to launch,” he said. “They’ve been saying publicly this fall, but in the meantime, there are a lot of [individuals] doing Omnipod DIY. If there is a tool to help diabetes be easier to manage with better numbers, that’s a first consideration.”
Patient Perspectives
Patients with T1D can spend hours managing their condition. This month, Drug Topics® sat down with Kate,* a New York City resident who received her T1D diagnosis at age 10, to share her perspective on the current state of diabetes technology.
Q: You’re in the process of making the change to an APS. What do you hope will change?
Kate: I’ve read online that the open-source algorithm designed by Open APS [Artificial Pancreas System] can control blood sugars a lot better than the algorithms currently in use.
I remember reading one testimonial from a teenage boy that the [Open APS] algorithm works so well, [he] doesn’t even tell the pump when he eats, and still keeps his blood sugars in relatively good range.
I’m hoping that once I am able to get all the necessary hardware for Open APS, I’ll have better managed blood sugars, especially after eating meals; that my blood sugars will be better managed when unexpected factors affect my blood sugars (such as stress); and that I will get better sleep.
Q: What do you want your health care providers to know about why you’re making this change?
One of the most frustrating things is knowing how much red tape is out there in terms of insurance coverage and companies being unwilling to work with the Open APS community to create the best product. I’m sure the commercial pump algorithms could work a lot better if they collaborated with the people working on this open- source algorithm.
T1D is a strange condition because it is so finicky and involves so much data management. If I didn’t have educated parents who were knowledgeable about medicine and technology to help me learn how to manage this condition, I would probably be in a much worse state than I am now. The best physicians I have had take the time to educate themselves and keep up to date on pump and sensor technology so they can advise me on what to get. I also think it’s important for diabetics to talk to an experienced diabetes educator regularly for help understanding patterns in their data. I love my endocrinologist, but she just doesn’t have time during appointments to go over that in detail.
Limits on insurance coverage are also problematic; some patents are restricted in terms of what’s covered and what’s compatible with Open APS. I’ve waited years for my insurance to approve an insulin pump from a different company that I think will likely work much better for me. But without insurance, insulin and supplies would be outrageously expensive. I have been scared to leave jobs, or even take a leap to start my own business, because I worry about insurance coverage for T1D. Health care providers are likely just as frustrated by these things as I am; it’s one of the most challenging aspects of managing diabetes in this country.
*Name has been changed
References
1. LoopDocs. Accessed May 20, 2022. https://loopkit.github.io/loopdocs/
2. The Nightscout Foundation. Accessed May 20, 2022. https://www.nightscoutfoundation.org/
3. OpenAPS.org. Accessed May 20, 2022. https://openaps.org/
4. Lewis D. Real-world use of Open Source Artificial Pancreas Systems—poster presented at American Diabetes Association Scientific Sessions. OpenAPS.org. Published June 11, 2016. Accessed May 20, 2022.
5. Tidepool. Accessed May 20, 2022. www.tidepool.org
6. Tidepool completes FDA 510(k) submission of Tidepool Loop. News release. Tidepool. January 7, 2021. Accessed May 20, 2022. https://www.businesswire.com/news/home/20210107005324/en/Tidepool-Completes-FDA-510-k-Submission-of-Tidepool-Loop
7. Look H. 99 years closer: About our FDA submission of Tidepool Loop. Tidepool. Published January 11, 2021. Accessed May 20, 2022. https://www.tidepool.org/blog/99-years-closer-about-our-fda-submission-of-tidepool-loop
8. Braune K, Lal RA, Petruzelkova L, et al; for the OPEN International Healthcare Professional Network and OPEN Legal Advisory Group. Open-source automated insulin delivery: international consensus statement and practical guidance for health-care professionals. Lancet Diabetes Endocrinol. 2022;10(1):58-74. doi:10.1016/S2213-8587(21)00267-9
9. Jeyaventhan R, Gallen G, Choudhary P, Hussain S. A real-world study of user characteristics, safety and efficacy of open-source closed-loop systems and Medtronic 670G. Diabetes Obes Metab. 2021;23(8):1989-1994. doi:10.1111/dom.14439
10. CamAPS® FX. CamDiab. Accessed May 20, 2022. https://camdiab.com/
